Unofficial BTS 3.13 patch

[Roland Johansen]

barbarian stealth DDs don't. it could be they're just stupid captains since they can't build them normally? i did a WB test just now, and barbs were the quickest way i could think of. i gave myself some stealth bombers and a stealth DD, the barbs got 2 stacks, each with a ship of the line and a stealth DD. bombing stack 1, the SoL was damaged to max and no further, but the SDD never admitted it was there. so it didn't take any damage and it never tried to hurt my planes. when my own SDD swam out and saw the barb SDDs, at that point the barb SDDs started reacting to my stealth bombers. after i officially legally knew they were there, they did take damage and they shot down one of my planes.

It's good that you tested this and it's pretty improbable that there exists a difference between barbarian units and non-barbarian units for aerial interception.

However, I'm a tiny bit worried about the size of your test sample: 2 stacks of 1 stealth destroyer and 1 ship of the line. The chance that an unmodded stealth destroyer intercepts units is 50% and stealth bombers have an evasion chance of 50%, so all in all a stealth destroyer only has a 25% chance to intercept a stealth bomber. And it doesn't take many stealth bomber missions to do maximum damage to a ship of the line, so the chance that none of these bombers is intercepted (even if aerial interception is working) is not insignificant.

I would test it myself, but I'm having computer problems which stops me from playing this great game until my computer is fixed.

By the way, the fact that the stealth destroyer was never attacked does seem to indicate that it doesn't intercept bombing missions. I just want to avoid the small chance that our reasoning about changes could be based on incorrect facts about how it works now.

I like this idea a lot.

There is one situation where it's not perfect: Say I have a Caravel with an Explorer and a Galleon with three Cannons. They get attacked by a Missile Cruiser. With your suggestion, the Galleon will defend because it has higher strength * odds and they both have cargo.
You're not distinguishing between different quantities of cargo.

I would guess that strength * odds ^ (#cargo / 2) would work pretty well.

Hmm, the more units a ship transports, the higher its perceived strength, the bigger the chance it will be used to defend. Hmm, interesting.

If at some point due to some code changes strong transport units with minimal units in their cargo bays (missile cruiser with 1 missile) are going to take precedence over weaker non-transport units (empty carrier), then the formula that governs this change should be very simple and documented in the civilopedia. I'm opposed to complicated formulas which are hidden to the user. There are enough of those already in the game.

I do like Wodan's proposal because it weakens the transport units so that they are only likely to defend when they will win the battle. However, it doesn't always produce the results that I would want and I would expect. Here is an example that is not unlikely to occur:

A galleon transporting 3 units (lets make them city raider 3 macemen to make you feel the pain ) and escorted by a caravel is attacked by an enemy caravel.

The galleon's chance of victory is 0.782 (78.2%) in this case. So it's likely to win, but victory is by no means a certainty.

Galleon strength according to Wodan: 4 * 0.782 = 3.128 which is higher than the strength 3 caravel so the galleon is going to defend. I would not want that. There is a significant chance that we lose 4 units. We want the caravel to defend and maybe even give its life to defend the galleon and its transported units.

I cannot criticise someone else's formulas without coming up with one that I think might work better. Of course, this is only my opinion, so if you can create scenarios where it isn't working as you would like it to, then please do say so.

The combat algorithms normally makes the unit which has the highest chance of victory defend. This also means that your expected losses are going to be minimized. That's just another way of describing the same rule. It is also exactly what you want when a stack of units containing transportation ships is going to be attacked. You want to minimize your losses.

When an empty galleon is going to be attacked by another galleon, then the chance of victory is 50% and the chance of loss is 50%, so on average you'll lose 0.5 units. If a galleon carrying 3 units is going to be attacked by another galleon, then the chances of victory remain the same, but the average amount of units that you'll lose in that battle is 0.5 * 4 = 2 because when you lose, you'll lose the galleon plus the 3 units it is carrying.

So my proposed rule is: calculate the expected number of losses ( = chance of loss * number of units that are going to be lost) and let the unit defend for whom this number is the lowest.

For example, the combat described above between the galleon + 3 transported units escorted by a caravel and the enemy caravel attacking.

Expected losses when the galleon is defending: 0.218 * 4 = 0.872
Expected losses when the caravel is defending: 0.5
So, the caravel will defend.

If the galleon were only carrying 1 unit, then it would defend:
Expected losses when the galleon is defending: 0.218 * 2 = 0.436
Expected losses when the caravel is defending: 0.5

By the way, I used this combat calculator to calculate the chances of victory and loss between various units.

 

[KMadCandy]

However, I'm a tiny bit worried about the size of your test sample: 2 stacks of 1 stealth destroyer and 1 ship of the line. The chance that an unmodded stealth destroyer intercepts units is 50% and stealth bombers have an evasion chance of 50%, so all in all a stealth destroyer only has a 25% chance to intercept a stealth bomber. And it doesn't take many stealth bomber missions to do maximum damage to a ship of the line, so the chance that none of these bombers is intercepted (even if aerial interception is working) is not insignificant.

I would test it myself, but I'm having computer problems which stops me from playing this great game until my computer is fixed.

By the way, the fact that the stealth destroyer was never attacked does seem to indicate that it doesn't intercept bombing missions. I just want to avoid the small chance that our reasoning about changes could be based on incorrect facts about how it works now.

and this is why i explain my HighlyScientificTestingMethods, so that people wiser than me can suggest ways to improve them . let's try it again, with airships and fighters and non-stealth bombers. the fact that WB lets you bypass the 4 air units per city limit really helps when you're doing this on a brand new map to not spoil surprise in your current game, btw *giggle*.

this time, i did get a message that a stealth destroyer shot down my airship. so you were completely right, thank you! once the SoL received max damage from my airships, the tile was greyed out for air strikes; they couldn't see the SDD to attack it. the fighters could attack the SoL again, since they had a higher damage limit to inflict on it. but once they hit their own cap, they were stuck too.

i guess that's intended. maybe the planes that get shot down never make it back to base to tell the others what happened to them? it just seems so weird.

Spoiler screenshot from before i sent my own stealth ship to see the barb stealh ship :

after i sent my own SDD out to find the barb ones, then my planes could hurt the barb SDDs. but first i had to see it with my own "this unit can see stealth units" eyes, not just infer it from the fact that pilots died.

each SDD only shot down one of my planes. was that coincidence, or are they like old SAM infantry, once they have one success they're done? i didn't think to test that until i'd loaded up my real game again, so i won't get to it any time soon

sidetrack: i was looking at the long list of units that barbarians cannot make today, and found out that privateers are not on the list. so apparently some of those privateers that look like barb privateers might actually be barb privateers! i just don't know quite what to think of that.

 

[mjs0]

So my proposed rule is: calculate the expected number of losses ( = chance of loss * number of units that are going to be lost) and let the unit defend for whom this number is the lowest.

Great post Roland!
I agree with most of what you said, just one comment...
Given your proposed rule, which I quoted above, why not go one step further and base the calculation on the total hammer value of the units rather than just the raw number of units. Using hammers would allow more granularity in the decision so that, for example, a transport carrying 4 modern armour would be far less likely to defend than one carrying 4 infantry.

Actually (as an aside)...I would like to try this approach (based on expected hammer loss) out in all combat situations not just naval combat involving loaded ships. Some adjustments might be necessary to assign a hammer value to XPs, promotions etc, but I think the results could be quite interesting.

 

[Refar]

I think a loaded unit should never defend. Yes it brings up some issues with the missile cruiser. But knowing it works this way, we can take the needed precautions.

Even a small chance of defeat is still a chance.

I have a fleet of loaded MC's and a couple of unloaded Transports ? I actually prefer loosing a unloaded transport over loosing (or even getting damaged) a Missile Cruiser. I can retaliate with those MC on the next turn. Transports are cheap bait.

When everyting is loaded the strongest shall defend.

 

[Roland Johansen]

and this is why i explain my HighlyScientificTestingMethods, so that people wiser than me can suggest ways to improve them . let's try it again, with airships and fighters and non-stealth bombers. the fact that WB lets you bypass the 4 air units per city limit really helps when you're doing this on a brand new map to not spoil surprise in your current game, btw *giggle*.

this time, i did get a message that a stealth destroyer shot down my airship. so you were completely right, thank you! once the SoL received max damage from my airships, the tile was greyed out for air strikes; they couldn't see the SDD to attack it. the fighters could attack the SoL again, since they had a higher damage limit to inflict on it. but once they hit their own cap, they were stuck too.

i guess that's intended. maybe the planes that get shot down never make it back to base to tell the others what happened to them? it just seems so weird.

Spoiler screenshot from before i sent my own stealth ship to see the barb stealh ship :

after i sent my own SDD out to find the barb ones, then my planes could hurt the barb SDDs. but first i had to see it with my own "this unit can see stealth units" eyes, not just infer it from the fact that pilots died.

each SDD only shot down one of my planes. was that coincidence, or are they like old SAM infantry, once they have one success they're done? i didn't think to test that until i'd loaded up my real game again, so i won't get to it any time soon

These kinds of mistakes are made all the time, even by people who claim that they understand probability and statistics. Thanks for the further testing, KmadCandy, the sneaky privateer.

The one interception per stealth destroyer isn't a coincidence. I did some testing a while ago because I thought that land based interception (Sam Infantry, Mobile Sam, etc.) would get more than one interception per turn in BTS and this made the ingame results that I observed a bit peculiar. I modded a Mobile Sam to 100% interception and added 20 other units to its tile, attacked it with 20 fighters and got exactly one interception. No coincidence, I'd say. (I assume here that sea interception works in the same way because it seems to be coded similar in xml.)

sidetrack: i was looking at the long list of units that barbarians cannot make today, and found out that privateers are not on the list. so apparently some of those privateers that look like barb privateers might actually be barb privateers! i just don't know quite what to think of that.

Another diplomatic explanation for the huge fleets of privateers that happen to appear on the seas shortly after your civilisation discovers the technology to build such ships? It's those cursed barbarians!

Great post Roland!

Thanks!

I agree with most of what you said, just one comment...
Given your proposed rule, which I quoted above, why not go one step further and base the calculation on the total hammer value of the units rather than just the raw number of units. Using hammers would allow more granularity in the decision so that, for example, a transport carrying 4 modern armour would be far less likely to defend than one carrying 4 infantry.

I actually thought about that way to do it too. There were two reasons why I chose not to use this rule.

1) It's more complicated while it is unlikely to make much of a difference in practical situations. While I do like complicated rule sets and nice formulas which rule the game, most people like formulas that they can understand without a lot of thought and even use during a game. If you start to use hammer values, then it becomes hard to calculate/predict (for a human) which unit will defend.

Civilization is mostly based on easy formulas which lead to complicated strategies. Adding details which don't have strategic results just complicates the game without adding to it.

2) The rules to pick the defender would then be different for transport type units and normal units. With my proposed rule, the defender is picked in a similar way for transport units and non-transport units. I personally like this, it makes the rule set more 'elegant'.

Actually (as an aside)...I would like to try this approach (based on expected hammer loss) out in all combat situations not just naval combat involving loaded ships. Some adjustments might be necessary to assign a hammer value to XPs, promotions etc, but I think the results could be quite interesting.

If you'd do that, then my objection 2 would be removed, it would make the defender rule set for non-transports and transports similar again.

You don't have to add hammer values to promotions. These promotions already affect the chance of victory and thus the expected loss.

But such a change would have all kinds of unexpected results. At present, unit costs are not solely based on the strength of the unit and the bonuses it has against certain units. It is also based on its movement rating and other special bonuses. That's the reason why cavalry units are more expensive. They can react more quickly to combat threats and thus be available in more situations increasing their worth. But that shouldn't make them hide behind other defenders which would happen when their higher cost is taken into account.

A second even better reason not to do this is because each unit has a value independent of its hammer cost. Each unit that you lose (in areas where you're not culturally dominant) causes extra war weariness. This war weariness is unrelated to the hammer cost of the unit. Making your high hammer cost units hide behind the cheaper defenders would result in higher war weariness.

A third reason would be that it would be more difficult for a player to predict which of his/her units is going to defend.

By the way, I like the way you think. I have similar ideas about this game, but I just think that a more complicated and detailed rule set doesn't necessarily lead to a better game. It might make it a better simulation, but not necessarily a better game.

 

[Roland Johansen]

I think a loaded unit should never defend. Yes it brings up some issues with the missile cruiser. But knowing it works this way, we can take the needed precautions.

Even a small chance of defeat is still a chance.

I have a fleet of loaded MC's and a couple of unloaded Transports ? I actually prefer loosing a unloaded transport over loosing (or even getting damaged) a Missile Cruiser. I can retaliate with those MC on the next turn. Transports are cheap bait.

When everyting is loaded the strongest shall defend.

I understand the way you think. With the suggestion I made a transport would defend before a fully loaded missile cruiser when attacked by a destroyer (chance of victory for the missile cruiser 78.2%, chance of victory for the transport 1.2%). When the missile cruiser wouldn't be fully loaded (3 or less missiles), then the missile cruiser would defend. This is based on the following situation:

5 stacks of 1 fully loaded missile cruiser and 1 empty transport attacked by 5 individual destroyers.

1) What would typically happen when the missile cruiser would defend:
The missile cruisers win 4 out of 5 fights and lose 1, so the losses are 1 missile cruiser and 4 missiles for you.

2) What would typically happen when the transport would defend:
The transports lose all of the fights, so the losses are 5 transports.

In situation 1, you also destroy 4 out of 5 destroyers. In situation 2, you might destroy them too if you can retaliate which is not always possible.

So both cases are remarkably close in outcome for the losses on your side. It's also clear that when the missile cruisers carry less missiles, that it becomes more and more attractive for them to defend instead of the transports. Most of us would prefer the missile cruiser to defend when it only carries 1 missile.

I don't necessarily want to convince you that I'm right or something like that. I just think that there is some merit to look at the losses that you might expect.

 

[Refar]

I see how your idea has some benefits. Minimizing the expected loss is basicly a good and valid approach. In any case it might be especially usefull as a 'suplementary' rule - when dealing with stacks of partially loaded units of the same type for example.

What makes me beeing not fully convinced is: While with "Empy unit allways defend" i know for sure what will be going on, with the more elaborate rule situation might arise, where i as player am not sure which of my units is going to defend in a upcomig fight. I might be too paranoid of "the computer is doing something 'behind my back'" here I might be not able to guess the odds of defending under influnece of partial damage, promotions...

Talking of promotions... My transports often get the Flanking/Navigation line - as far as i have the Exp to go, considering they are usually not built in primary military cities.

So maybe the rule should say "Chance of loss and not retreat * number units"

 

[Roland Johansen]

I see how your idea has some benefits. Minimizing the expected loss is basicly a good and valid approach. In any case it might be especially usefull as a 'suplementary' rule - when dealing with stacks of partially loaded units of the same type for example.

What makes me beeing not fully convinced is: While with "Empy unit allways defend" i know for sure what will be going on, with the more elaborate rule situation might arise, where i as player am not sure which of my units is going to defend in a upcomig fight. I might be too paranoid of "the computer is doing something 'behind my back'" here I might be not able to guess the odds of defending under influnece of partial damage, promotions...

Talking of promotions... My transports often get the Flanking/Navigation line - as far as i have the Exp to go, considering they are usually not built in primary military cities.

So maybe the rule should say "Chance of loss and not retreat * number units"

This rule would be a bit more complicated than the present 'transport units do not defend until all non-transport units have been defeated' and so yes there would be situations where it might be hard to guess which unit will defend especially when units are getting wounded. I do think it would often choose the unit that you might intuitively prefer yourself because most players would like to lose as few units as possible. And I also think that the civilopedia should be updated when such a rule change would be made.

Defending units cannot ever retreat, so this should not be an issue. But if they could retreat, then you're right that that would be the better rule.

 

[Refar]

Defending units cannot ever retreat, so this should not be an issue. But if they could retreat, then you're right that that would be the better rule.

I seen those loading screen tooltip "Flanking only works when you attack" line 10000 times now. So basicly i knew it. I was still giving flanking to my transports in every single game Well at least Navigatino was of some use.

 

[Roland Johansen]

I seen those loading screen tooltip "Flanking only works when you attack" line 10000 times now. So basicly i knew it. I was still giving flanking to my transports in every single game Well at least Navigatino was of some use.

Oh, I do that too. I give transports the navigation upgrade because transports are the units with the lowest movement rate in an invasion fleet. And the way to get to the navigation upgrade it through flanking.

 

[Aquatic]

While I hate to pour cold water on the excellent suggestions regarding the cargo/escort issue (all of which deserve some thought, though personally I'm still slightly more in favour of the status quo), isn't the current rule of 'transport units do not defend until all non-transport units have been defeated' actually functioning as designed? Isn't this therefore a matter to be considered for a mod, or that perhaps could be referred to Firaxis for a rule change, but not something that should be part of the 'unofficial patch' - which is not the case with the issue of stealth units refusing to defend when on escort duty?

None of which is to say that I have not enjoyed the discussion! We just need as much as possible to avoid straying from patching to modding (not an easy line to walk sometimes, as there are some very grey areas...).

 

[jkp1187]

We just need as much as possible to avoid straying from patching to modding (not an easy line to walk sometimes, as there are some very grey areas...).

After sorting through the threads and trying to get my mind around some of the numbers, I'm starting to agree with Aquatic and Refar here. This might be one of those cases where the rule might be iffy, but even the lest-optimal rule, if it's objective and clear, is better than an 'optimal' rule that is less easily understandable. Players can plan around it -- better to send two unloaded missile cruisers and two fully-loaded missile cruisers in a stack than four Missile Cruisers with two missiles each. If there's one thing players don't like (at least, one thing this player doesn't like) is complete uncertainty in how the computer is going to select the defending units.

I would be very upset if the computer would select, for instance, an undamaged aircraft carrier with three fighters to defend over a heavily damaged transports. After all, sometimes it's better for the weakest units that are damaged and guaranteed to lose to be the defending units. If the computer is forced to attack every unloaded transport before the cargo carriers step up to defend, if nothing else, the sacrificial lamb units have been absorbing attacks, and the enemy won't be able to use those attacking units to attack anything else that turn (except in the unlikely event that the enemy's units have Blitz, which it's kind of hard for naval units to get b/c it requires Combat III...and even if they do have Blitz, those units will be at least a little more damaged, and thus perhaps less likely to successfully take on a full-strength missile cruiser, for example.) In short: let 'em sink my empty transports, and I can counterattack the next turn with my full-strength units.

So perhaps the best thing that could be done for newbie players is to add another entry to the Concepts section of the Civilopedia explaining exactly how the computer chooses defenders in a stack.

 

[KMadCandy]

So perhaps the best thing that could be done for newbie players is to add another entry to the Concepts section of the Civilopedia explaining exactly how the computer chooses defenders in a stack.

i really really like that idea! i just re-read the Combat entry under Game Concepts and it's not even hinted at there. get thee to the Suggestions sub-forum! if i posted it, i'd use way too many words .

 

[Duuk]

Can anything be done to fix flanking? Seriously, knights should not be able to kill all the cannons while the riflemen sit and watch.

Perhaps make siege units ALWAYS defend last if they are in a stack?

 

[ori]

Can anything be done to fix flanking? Seriously, knights should not be able to kill all the cannons while the riflemen sit and watch.

Perhaps make siege units ALWAYS defend last if they are in a stack?

While you may not like it it is a deliberate change made by the designers in BtS so I would not regard it as a bug...

 

[SevenSpirits]

>I would guess that strength * odds ^ (#cargo / 2) would work pretty well.

Hmm, the more units a ship transports, the higher its perceived strength, the bigger the chance it will be used to defend. Hmm, interesting.

I'm not sure if you actually misunderstood me, but the odds are always less than one, and exponentiation has higher precedence than multiplication. If you're getting higher numbers by increasing the cargo you're doing something wrong.

 

[Jaybe]

.... This might be one of those cases where the rule might be iffy, but even the lest-optimal rule, if it's objective and clear, is better than an 'optimal' rule that is less easily understandable. Players can plan around it -- ....

I would be very upset if the computer would select, for instance, an undamaged aircraft carrier with three fighters to defend over a heavily damaged transports.

NO! NO! A Thousand Times NO! Predictability is NOT Good. Especially in warfare, surprises HAPPEN.

If aircraft had separate damage capabilities between ground and naval units (so you could sink ships without being able to destroy ground units), then with unpredictability you could have a Midway -- go straight for the carriers. You think those flyboys are going to go after the ESCORTS??!!!

Yes, it should be DOCUMENTED, but there should be a 1/4 chance of NOT having to fight the strongest unit; along with other "it will never happen in civ" concepts of ...
defending units should be able to retreat,
after using siege units for bombardment and reducing the defending tile, there should be just ONE (well, maybe two) attacks for the stack -- one grand battle, not all this skirmishing to allow for breaking off the assault if it's not going well.

--
Regarding unsighted stealth destroyers defending against naval units: they should defend as normal even if they are not knowingly there; those that do actually defend would be revealed.

Re loaded & unloaded naval units defending, all stated options are good, including leaving it as is (bleepin' cowards leaving those transports defenseless -- or were the attackers devious enough to distract the escorts)?

 

[jkp1187]

1.) Civ != wargame. Here are some wargames if you're interested.

2.) Don't get me wrong, the calculations discussed above are not bad, in concept, and personally I wouldn't mind a mod that added a bit more complexity to combat resolutions, but I don't think the suggested changes belong in a patch that is trying to be as faithful to the designers' intent as possible. And it would have to be a mod.

 

[Roland Johansen]

I'm not sure if you actually misunderstood me, but the odds are always less than one, and exponentiation has higher precedence than multiplication. If you're getting higher numbers by increasing the cargo you're doing something wrong.

Okay, that was maybe the stupidest thing I've ever written. Sorry. I shouldn't post directly after waking up. Hiding in shame.

( I thought to have read (strength * odds) ^ (# cargo/2 ) ).

 

[Minor Annoyance]

Can anything be done to fix flanking? Seriously, knights should not be able to kill all the cannons while the riflemen sit and watch.

Perhaps make siege units ALWAYS defend last if they are in a stack?

I'm not sure I understand. Knights can't flank attack a cannon, only Catapults and Trebuchet. Even if it was a Cuirassier against a cannon, they'd still be fighting the riflemen and they'd have to win or retreat to damage the siege units, and that damage if applied without the siege units actually being the defender so It wouldn't matter if they were made to always be the last to defend because in the situation you outlined they wouldn't be defending anyway.